> daily_signal(2026_05_10)

A Michigan farm town voted no on a $16 billion AI data center. The developer sued. Construction broke ground anyway.

PickBits Daily Signal · Sunday, May 10, 2026

By Mark Pickering · 6 min read · May 10, 2026

// tl;dr

This week, Saline Township, Michigan, learned its 4-1 vote against a $16 billion OpenAI/Oracle data center on 1,000 acres of local farmland did not matter. The developer sued, the township settled, and construction had already broken ground. Reported in Fortune on May 6. The same week, the chip in your next laptop might be moving from Taiwan to Arizona because Trump and Lutnick personally pushed it. Your Instagram follower count drop is the visible side of Meta's first-ever decline in daily users. Every AI tool your team uses at work turns out to share one shareholder: NVIDIA's $40 billion in investments cycles right back as chip purchases. And the International Monetary Fund said that an AI-powered cyberattack on the banking system is now inevitable.

When a town can vote no on a $16 billion data center and still have it built, the decision was never theirs to make. The day-to-day choices about your power grid, your farmland, and the chip in your laptop are being made for you, not by you.

1. A Michigan farm town voted no on a $16 billion AI data center. The developer sued. Construction broke ground anyway.

In September, the township board of Saline Township, Michigan (a farming community of about 2,800 people with an annual budget of roughly $1 million) voted 4-1 to deny rezoning for a planned 21-million-square-foot data center on 1,000 acres of farmland. The planning commission had already rejected it. The data center, a $16 billion project for the OpenAI/Oracle Stargate initiative, would draw 1.4 gigawatts of electricity from the regional grid, comparable to a nuclear plant. Developer Related Digital (a subsidiary of Related Companies, founded by billionaire Stephen Ross) sued the township within 24 hours under a legal doctrine called exclusionary zoning, which argues that local zoning unreasonably blocks legitimate land use. Faced with a lawsuit it could not afford to fight, the township settled within weeks. Construction broke ground in November. Related Digital announced in April that the $16 billion in project financing had closed. Fortune's Sharon Goldman reported the full timeline in a May 6 piece.

The settlement: roughly $14 million in community benefits, or about 14 times Saline Township's annual budget. About $4 million of that went to farmland-preservation easements; the rest covered noise caps, water-use limits, and expansion restrictions. Township attorney Fred Lucas told Fortune the township was "between a rock and a hard place." Resident Kathryn Haushalter, a former Marine and mother of five: "I feel like I'm playing by a different rule book." Opposition group member Joshua LeBaron: "An AI stock market crash is probably the only thing that could stop it now." This is not an isolated Michigan story. The same week the Fortune piece dropped, Utah's Stratos project hit a 21-outlet coverage backlash, and Governor Spencer Cox moved to make state-level approvals harder; a man was body-slammed at a Chesterfield Township meeting in New Jersey over a data-center vote; Ireland's Echelon got hit with a €2.67 million Wicklow County Council bill on its Arklow project; and tax-credit pushback intensified across Upstate New York. Saline is the case where the legal playbook was tested.

Fortune coverage of Saline Township Michigan voting against the OpenAI/Oracle Stargate data center
fortune.com · May 6, 2026
Why this matters: The lawsuit Related Digital filed under exclusionary zoning is a tested legal playbook that says local governments cannot reasonably block legitimate land uses. Tested in Saline, the doctrine produced a $14 million settlement for a township with a $1 million annual budget. The lesson every developer learns: sue the small towns. They cannot afford to fight. Action this week: Find your county or township's planning-commission agenda page online and sign up for email alerts. Read AI Now's free policy toolkit at datacenters.ainowinstitute.org/local, which lists protective zoning language from real US towns you can hand to your board. Show up before a proposal hits the agenda. Once it does, you are on the developer's timeline.

https://fortune.com/2026/05/06/ai-data-center-michigan-saline-politics-farmland/

2. The chip in your next laptop might be made in Arizona. The White House personally pushed it there.

The chip in your next laptop might be made in Arizona, Ohio, or Ireland instead of Taiwan. That is what the Apple-Intel deal means in practice, the Wall Street Journal reported on May 8. Apple agreed in principle to have Intel make the future processors for Macs and iPads, with production possibly starting in 2027. Intel shares closed up roughly 15 percent on the news; Apple rose about 1.7 percent. The companies had been in talks for more than a year. The reason any of this is happening: the Journal reported that Commerce Secretary Howard Lutnick met repeatedly over the past year with Tim Cook, Elon Musk, and Jensen Huang to push them to put silicon work through Intel, and that President Trump personally raised the topic with Cook in a White House meeting.

This is the first time in a generation that the federal government has visibly rerouted where consumer electronics get made. Apple has been a Taiwan-Semi customer for nearly a decade. That does not stop. But a slice of Apple's high-end chip-making demand moves to US soil, onto an Intel plant that had been struggling to land a named first customer for its newest manufacturing lines. The price the federal side appears to have paid is the 10 percent stake the US government took in Intel earlier this year. NVIDIA was reported in March to have taken a $5 billion position in Intel, and Lutnick has been making the same rounds with Musk and Huang. Watch which household-name device gets the next reveal.

CNBC coverage of the Apple-Intel preliminary chip-making deal
cnbc.com · May 8, 2026
Why this matters: If you've bought a Mac or an iPhone in the last decade, the chip inside came from Taiwan. Starting around 2027, Apple says some of those chips might instead come from Intel in the United States. That happened because the federal government took a 10 percent stake in Intel earlier this year and personally lobbied Tim Cook to use them. Action this week: if you're buying a Mac or iPad in the next 12 months, check the "Designed by Apple in California, assembled in..." line on the box before you pay — by 2027 Apple silicon ship-from could split between TSMC Arizona (Intel-fabbed) and TSMC Taiwan, and the resale and warranty-replacement supply differs. If you're an IT buyer or hardware architect at work, pull your three-year refresh roadmap and flag any line item that assumes single-source-from-Taiwan chips (Mac fleets, GPU-heavy workstations, embedded silicon for any TSMC-only design); ask your reseller for written confirmation on fab-of-origin and US-vs-Taiwan stock allocation before signing the next PO. If you hold Intel in a 401(k) or brokerage account, the named-customer milestone is the catalyst — set a watch alert for the formal Apple announcement and which Intel manufacturing line gets named.

https://www.cnbc.com/2026/05/08/intel-stock-apple-chip-deal.html

3. Your Instagram follower count drop wasn't your imagination. Meta lost daily users for the first time ever.

If you have noticed your Instagram follower count drop over the past several weeks, you are part of two things at once. The first: Meta ran a cleanup that deleted millions of accounts (Meta describes them as fake or inauthentic; users with sudden follower drops told reporters their accounts were not). The second, reported on Q1 2026 earnings and revisited in New York Times and Futurism coverage this week: the total number of people using Meta apps every day fell for the first time ever, by roughly 20 million, to 3.56 billion. CEO Mark Zuckerberg blamed internet disruptions in Iran and a WhatsApp restriction in Russia. Analysts were skeptical. The same earnings raised 2026 AI infrastructure spending guidance to a range of $125 billion to $145 billion, up $10 billion. Revenue rose 33 percent to $56.31 billion. The stock fell 7 percent. Meta laid off 8,000 in the same quarter. On May 9, Julia Angwin in the New York Times called the combined picture a "death spiral": rising spend, falling user base, an AI bet that has not produced consumer pull.

This matters for anyone who runs a brand page, a creator account, a community, or just keeps in touch with people via Instagram or WhatsApp. The platform you spent years building reach on is shrinking for the first time. Meta's defense for spending $145 billion on AI used to be "we run the largest distribution surface in the world." That story is harder to hold when the surface is contracting. The follow-on question for users is more practical: which of your friends and family are still actually using Facebook or Instagram every day, and where did the rest of them go?

The Next Web coverage of Meta Q1 2026 user decline against rising AI spending
thenextweb.com · May 2026
Why this matters: If you use Instagram, Facebook, WhatsApp, or Threads, Meta just said the total number of people on their apps every day went down for the first time ever. About 20 million people are gone. In the same quarter, they raised what they're spending on AI infrastructure to a range of 125 to 145 billion dollars. The audience is shrinking while the spending goes up. Action this week: Go to accountscenter.facebook.com and download a full zip of your Meta accounts before another sweep hits. Get phone numbers for the people you only talk to in Messenger or WhatsApp, so you can reach them if your account or theirs disappears. If you run a brand page or creator account, start moving people to your own email list this week (Beehiiv, Substack, or ConvertKit will import what you have). If your follower count dropped overnight in the cleanup, file an appeal through Meta's account help.

https://thenextweb.com/news/meta-q1-2026-user-decline-ai-capex

Tomorrow's signal lands here. Subscribe to PickBits Daily Signal if you want the daily that leads with impact, not hype.

4. Every AI tool your team uses at work is bankrolled by the same company. NVIDIA disclosed $40 billion of the loop.

The AI tools your team uses at work (ChatGPT for emails, Claude for writing, Copilot for code, even the AI features built into your software) were almost all made by companies that have NVIDIA money inside them. NVIDIA does not just sell the chips. It invests in the companies that buy the chips. On May 9, CNBC, TechCrunch, and Benzinga reported that NVIDIA's publicly disclosed AI investments for the year so far crossed $40 billion. The single largest position: $30 billion into OpenAI. Beyond OpenAI: up to $3.2 billion in Corning; up to $2.1 billion in data-center company IREN, which signed a separate $3.4 billion contract to buy NVIDIA chips earlier this month. Plus stakes in roughly two dozen smaller AI companies. Wayve, Mistral AI, Lambda Labs, Genesis Therapeutics, Recraft, JetBrains.

The dollars don't move outward. They go in a circle. NVIDIA invests in OpenAI. OpenAI uses the money to buy NVIDIA chips. The money is back at NVIDIA. Same with IREN: NVIDIA invested $2.1 billion, and IREN signed a $3.4 billion chip-buying contract. Wall Street analyst Matthew Bryson at Wedbush called the deals "squarely into the circular investment theme." For anyone using AI tools at work, the practical read is that the five companies behind them are tied together at the bank. If NVIDIA hits a bad quarter, every AI tool you use (and the prices they charge) move at the same time. The number to watch is not $40 billion. It is how much of NVIDIA's revenue this year is being paid for by NVIDIA's own balance sheet, and what happens to the tools you use if the loop breaks.

CNBC coverage of NVIDIA $40B AI equity portfolio in 2026
cnbc.com · May 9, 2026
Why this matters: If you've used ChatGPT or Claude or any AI tool, the company that made it probably has NVIDIA money in it. NVIDIA disclosed 40 billion dollars of investments this year, with 30 billion going into OpenAI alone. That money flows back to NVIDIA when those companies buy chips. Action this week: open your company's AI vendor list (or your own personal AI subscriptions in Stripe/your bank) and mark every line where NVIDIA is on the cap table — OpenAI (ChatGPT, Copilot, the model under most "AI features"), Mistral, Lambda Labs, Recraft, JetBrains, Wayve, IREN, Genesis Therapeutics. That's your concentration map. If you sign AI vendor contracts at work, pull the active SaaS agreements and search for "price adjustment," "compute cost pass-through," or "infrastructure escalator" clauses; the ones tied to GPU pricing are the ones that move if NVIDIA hits a bad quarter — get a fixed-rate carve-out at renewal. If you hold NVDA in any brokerage account, the disclosure to watch on the next quarterly is whether NVIDIA breaks out chip-sales revenue from customers it has also invested in (Cisco around 1999 had to). If you advise clients on AI strategy, the read for them is the same: pricing risk is correlated across "different" vendors that share one funder.

https://www.cnbc.com/2026/05/09/nvidia-embraces-ai-investor-topping-40-billion-in-equity-bets-2026.html

5. If a hard conversation with ChatGPT flags a crisis, it can now alert a person you trust.

If you have ever had a hard conversation with ChatGPT. Venting about something serious, asking for help in a crisis. OpenAI shipped a feature on May 9 that changes what happens next. The Trusted Contact rollout lets a distress-flagged conversation alert a person you have designated in advance, not just put a hotline number on the screen. That is a real change in what a chatbot can do in the worst moment of someone's day.

OpenAI's walkthrough on how to set it up covers the enrollment flow, the age and region rules, what your trusted contact actually receives (a short alert, not the conversation), and how to change or remove the contact. Worth opening today if you, or someone in your house, uses ChatGPT for personal stuff.

For anyone using ChatGPT in daily life, this is a real escalation path that now exists where there used to be only a phone number.

ChatGPT Trusted Contact safety feature alerts a designated person if a conversation flags a crisis
euroweeklynews.com · May 9, 2026
Why this matters: If you talk to ChatGPT, it can now alert a person you trust if a conversation flags a crisis. Action this week: open OpenAI's setup page and add the person you would actually want to hear from in a hard moment. Tell them you did, so they accept the invitation when it arrives. The feature stays off until both sides confirm.

https://euroweeklynews.com/2026/05/09/chatgpt-can-now-warn-a-friend-or-family-member-if-it-believes-a-user-may-be-in-danger/
https://openai.com/index/introducing-trusted-contact-in-chatgpt/

Δ The counter-signal - the IMF said an AI-powered cyber attack on your bank is now inevitable.

If you bank with anyone (your local credit union, a national bank, a fintech app on your phone), the International Monetary Fund just said the people behind your bank account are now exposed to an AI-powered cyber attack at a scale that could hit a lot of banks at once. The IMF's May 7 financial-stability blog post, titled "Financial Stability Risks Mount as Artificial Intelligence Fuels Cyberattacks," argued that advanced AI lowers the cost and time required to find and exploit software bugs, and the fact that the global banking system runs on a small handful of the same cloud providers and AI vendors means one exploited weakness could cascade across many banks at the same time. Extreme cyber-attack losses, the IMF said, could "trigger funding strains, raise solvency concerns, and disrupt broader markets." The post highlighted Anthropic's controlled Mythos release as evidence that the threat is outpacing the rules. The same week, the Federal Reserve and the Treasury convened bank CEOs on cyber risk.

The previous five stories are about who is winning the AI build-out: a Michigan town that voted no but got built on, a White-House-pushed chip-supply pivot, AI spending defended against a falling user count, a $40 billion in-a-circle financing chain, and a new ChatGPT safety feature. The IMF report is that the attack surface those tools depend on is widening faster than any single fix can close it. The same handful of cloud providers. The same handful of chip suppliers. The same handful of AI vendors. The IMF named the concentration directly. So did the Saline lawsuit, in its own way: the rules that are supposed to constrain this are being outrun.

IMF blog post on AI fuelling cyberattacks against financial system
imf.org · May 7, 2026
Why this matters: If your bank, hospital, insurance company, or anyone else holding your data uses AI or runs in the cloud (and they all do), the IMF just said this is becoming a financial-stability risk. AI makes hacks faster and cheaper, and almost everyone uses the same handful of cloud providers and AI vendors. One breach can hit many companies at once. Action this week: log into your bank's secure-message portal (or call the relationship line if you have one) and ask which cloud provider and which AI vendors they rely on; if they can't answer, the answer is AWS, Azure, GCP, and one or two of OpenAI/Anthropic/Microsoft — write the answer down and use it the next time you're rebalancing your deposit concentration across institutions. If you run IT or security at work, pull your top 3 vendor list and run a concentration audit this week: which of your SaaS tools are on the same hyperscaler, which share an AI subprocessor, and which authentication providers sit in front of all of them; the deliverable is a one-page heat map your CFO can read. If you're an analyst or BA on a financial-services account, subscribe to the IMF Financial Stability blog and the Federal Reserve press releases — the May 7 IMF post is the framing rules will pick up first, so the proposed-rule language on vendor diversification is the next shoe to drop.

https://www.imf.org/en/blogs/articles/2026/05/07/financial-stability-risks-mount-as-artificial-intelligence-fuels-cyberattacks

» What to watch for

Tomorrow's signal lands here.